Skip to content
Evidence A2 min read·Last reviewed 2026-05-17

Sleep stages explained: NREM, REM, and what cycles look like

What the four stages of sleep represent, how cycles change across the night, and what consumer wearables can and can't tell you about them.

Reviewed by The Biohacking Bible editorial team

Bookmark

Sleep is not a uniform off-state. The AASM scoring guidelines divide it into four stages, defined by EEG, eye movement (EOG), and muscle tone (EMG) patterns measured during polysomnography.

Schematic hypnogram across eight hours. Slow-wave (N3) sleep concentrates in the first third; REM episodes lengthen and concentrate in the last third.

The four stages#

N1: light transition. A short stage at sleep onset; only a few minutes typically. Easy to wake from.

N2 — light sleep. The majority of total sleep time in healthy adults — usually 45–55%. Characterised by sleep spindles and K-complexes on EEG. Associated with memory consolidation in a number of studies.

N3: slow-wave (deep) sleep. EEG dominated by slow delta waves. Concentrated in the first third of the night. This is the stage most associated with growth hormone release, glymphatic clearance research, and the subjective sense of "deep" rest.

REM: rapid eye movement. Distinctive EEG that resembles waking, paralysed skeletal muscles, and the eye movements that give the stage its name. Concentrated in the last third of the night. Most vivid dreaming occurs in REM. Associated with emotional memory and procedural learning.

What a cycle looks like#

A typical adult cycle is around 90 minutes and runs N1 → N2 → N3 → N2 → REM. Across a night you get four to six cycles. Slow-wave sleep dominates the early cycles; REM episodes get longer and more frequent as the night progresses, which is why the second half of the night is rich in REM and the first half is rich in deep sleep.

How proportions change with age#

Slow-wave sleep declines from a peak in childhood through adolescence and adulthood, especially in men. REM proportion is relatively preserved into older age but the total amount drops with total sleep time. Older adults typically have more fragmented sleep with more brief awakenings.

What consumer wearables can and can't measure#

Wrist-worn devices estimate sleep stages from movement and heart-rate patterns (and sometimes wrist temperature or SpO2). Validation studies against polysomnography show:

  • Two-class accuracy (sleep vs wake) is generally good.
  • Four-class staging (light / deep / REM / wake) is substantially less accurate, with REM and deep often confused and stage transitions mis-timed.
  • Trends over weeks within an individual are more reliable than single-night absolute values.

For practical purposes, treat a consumer wearable's "deep sleep" number as a noisy proxy, useful for tracking your own trend but not for comparing against a friend's device. The actual research on what slow-wave or REM percentages should look like is based on PSG, not wearable estimates.

Practical reads#

If you only remember three things: most cycles are about 90 minutes, deep sleep concentrates early, and REM concentrates late. Cutting an hour off either end of your night has a disproportionate effect on the kind of sleep you lose.

Frequently asked

If my watch says I got 40 minutes of deep sleep, is that good?
Treat the absolute number with skepticism — wearables agree with PSG on total sleep time but mis-classify stages frequently. The within-individual trend across weeks is more reliable than a single night's number.
Why do I remember dreams more in the morning?
REM episodes get longer toward the second half of the night, so morning awakening is more likely to interrupt REM — and you remember the dream from the moments before waking.
Can I train my body to need less sleep?
No clean evidence supports this. Trials of chronic restriction show cumulative cognitive and metabolic costs even when subjects report feeling adapted.

References

  1. Patel AK et al. (2022). Physiology, Sleep Stages. StatPearls / NCBI Bookshelf
  2. de Zambotti M et al. (2019). The sleep of the ring: Comparison of the OURA sleep tracker against polysomnography. Behavioral Sleep Medicine
  3. Carskadon MA, Dement WC (2017). Normal human sleep: An overview (Principles and Practice of Sleep Medicine, 6th ed). Elsevier

Was this page useful?

Votes are pseudonymous. No account, cookie, or email needed. We aggregate feedback weekly to prioritise page reviews.

Related protocols

Last reviewed: · Evidence grade: A

Reviews check that sources are still current, that nothing has been recalled or had its MHRA status changed, and that no banned claims have crept in. See our grading methodology and corrections policy.